style_transfer1.py

# https://deeplearningcourses.com/c/advanced-computer-vision
# https://www.udemy.com/advanced-computer-vision

from __future__ import print_function, division
from builtins import range, input
# Note: you may need to update your version of future
# sudo pip install -U future

# In this script, we will focus on generating the content
# E.g. given an image, can we recreate the same image

from keras.layers import Input, Lambda, Dense, Flatten
from keras.layers import AveragePooling2D, MaxPooling2D
from keras.layers.convolutional import Conv2D
from keras.models import Model, Sequential
from keras.applications.vgg16 import VGG16
from keras.applications.vgg16 import preprocess_input
from keras.preprocessing import image

import keras.backend as K
import numpy as np
import matplotlib.pyplot as plt

from scipy.optimize import fmin_l_bfgs_b



def VGG16_AvgPool(shape):
  # we want to account for features across the entire image
  # so get rid of the maxpool which throws away information
  vgg = VGG16(input_shape=shape, weights='imagenet', include_top=False)

  new_model = Sequential()
  for layer in vgg.layers:
    if layer.__class__ == MaxPooling2D:
      # replace it with average pooling
      new_model.add(AveragePooling2D())
    else:
      new_model.add(layer)

  return new_model

def VGG16_AvgPool_CutOff(shape, num_convs):
  # there are 13 convolutions in total
  # we can pick any of them as the "output"
  # of our content model

  if num_convs < 1 or num_convs > 13:
    print("num_convs must be in the range [1, 13]")
    return None

  model = VGG16_AvgPool(shape)
  new_model = Sequential()
  n = 0
  for layer in model.layers:
    if layer.__class__ == Conv2D:
      n += 1
    new_model.add(layer)
    if n >= num_convs:
      break

  return new_model


def unpreprocess(img):
  img[..., 0] += 103.939
  img[..., 1] += 116.779
  img[..., 2] += 126.68
  img = img[..., ::-1]
  return img


def scale_img(x):
  x = x - x.min()
  x = x / x.max()
  return x


if __name__ == '__main__':

  # open an image
  # feel free to try your own
  # path = '../large_files/caltech101/101_ObjectCategories/elephant/image_0002.jpg'
  path = 'content/elephant.jpg'
  img = image.load_img(path)

  # convert image to array and preprocess for vgg
  x = image.img_to_array(img)
  x = np.expand_dims(x, axis=0)
  x = preprocess_input(x)

  # we'll use this throughout the rest of the script
  batch_shape = x.shape
  shape = x.shape[1:]

  # see the image
  # plt.imshow(img)
  # plt.show()


  # make a content model
  # try different cutoffs to see the images that result
  content_model = VGG16_AvgPool_CutOff(shape, 11)

  # make the target
  target = K.variable(content_model.predict(x))


  # try to match the image

  # define our loss in keras
  loss = K.mean(K.square(target - content_model.output))

  # gradients which are needed by the optimizer
  grads = K.gradients(loss, content_model.input)

  # just like theano.function
  get_loss_and_grads = K.function(
    inputs=[content_model.input],
    outputs=[loss] + grads
  )


  def get_loss_and_grads_wrapper(x_vec):
    # scipy's minimizer allows us to pass back
    # function value f(x) and its gradient f'(x)
    # simultaneously, rather than using the fprime arg
    #
    # we cannot use get_loss_and_grads() directly
    # input to minimizer func must be a 1-D array
    # input to get_loss_and_grads must be [batch_of_images]
    #
    # gradient must also be a 1-D array
    # and both loss and gradient must be np.float64
    # will get an error otherwise

    l, g = get_loss_and_grads([x_vec.reshape(*batch_shape)])
    return l.astype(np.float64), g.flatten().astype(np.float64)



  from datetime import datetime
  t0 = datetime.now()
  losses = []
  x = np.random.randn(np.prod(batch_shape))
  for i in range(10):
    x, l, _ = fmin_l_bfgs_b(
      func=get_loss_and_grads_wrapper,
      x0=x,
      # bounds=[[-127, 127]]*len(x.flatten()),
      maxfun=20
    )
    x = np.clip(x, -127, 127)
    # print("min:", x.min(), "max:", x.max())
    print("iter=%s, loss=%s" % (i, l))
    losses.append(l)

  print("duration:", datetime.now() - t0)
  plt.plot(losses)
  plt.show()

  newimg = x.reshape(*batch_shape)
  final_img = unpreprocess(newimg)


  plt.imshow(scale_img(final_img[0]))
  plt.show()